Accurate compliance control method for grinding and polishing robot under uncertain environment

Abstract

The invention relates to the technical field of industrial robot intelligent control, and discloses a precise compliance force control method of a grinding and polishing robot under an uncertain environment, and the method comprises the steps: building an environment-mechanical arm double-closed-loop control model according to a set linear environment model and a mechanical arm dynamics model; for an environment characteristic unknown part in the control model, iteratively solving an optimal control gain in a data driving mode based on adaptive dynamic programming; designing a free space control gain to realize a stable relative motion approaching environment; designing a free-contact space control gain smooth transition mechanism; model-free optimal control based on data iteration is achieved through self-adaptive dynamic programming, the impedance effect of force is achieved through double-closed-loop impedance control, modeling is conducted through contact transient impact, control gain can be dynamically adjusted, the stability of contact force and the machining quality are guaranteed, and the method is suitable for various complex working environments.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control of industrial robots, and more particularly to a precise compliant force control method of a grinding and polishing robot under environmental uncertainty. BACKGROUND

[0002] With the rapid progress of technology and the intensification of global manufacturing competition, industrial robots play an increasingly important role in precision manufacturing processes. Especially in high-precision processing scenarios such as grinding and polishing, industrial robots gradually replace traditional manual operation modes by achieving precise operation and stable quality assurance, greatly improving production efficiency and product quality. Not only does it meet the development trend of "intelligent manufacturing", but it is also a key entry point for the intelligent upgrading of manufacturing and the innovation of industrial robot technology. In the fields of aerospace, marine equipment and automobile manufacturing, precision grinding and polishing operations have extremely strict requirements for surface smoothness and shape accuracy. However, due to the complex and variable grinding and polishing operation environment and multiple error sources, existing grinding and polishing industrial robots are difficult to fully meet the high standards of product consistency and efficiency required by modern industry. Therefore, the demand for precise compliant force control of grinding and polishing industrial robots under environmental uncertainty is increasingly prominent.

[0003] In the field of compliant force control of industrial robots, traditional sensor-based force / dynamic control methods (such as impedance control and admittance control) are widely used to handle the interaction between the robot and the environment. Among them, impedance control models the dynamic relationship between the end pose and the contact force as a mass-spring-damper system, providing an easily adjustable dynamic response mechanism. However, traditional impedance control often has difficulty achieving precise force control when facing unknown or time-varying environmental stiffness, and is prone to contact force overshoot or transient instability. In grinding and polishing tasks, this control defect may cause damage to the material surface or degrade the processing quality, seriously affecting production efficiency and product consistency.

[0004] In order to solve the above problems, various improvement strategies have been proposed in recent years, including methods based on trajectory deformation and adaptive impedance parameter adjustment. The trajectory deformation method adjusts the reference motion trajectory dynamically to adapt to unknown environmental stiffness, however, the control bandwidth of this method is limited, making it difficult to cope with scenarios where environmental stiffness changes rapidly. In contrast, the adaptive impedance parameter adjustment method updates the impedance parameters (such as stiffness and damping coefficients) online to adapt to environmental characteristics, but these methods usually require a predefined set of offline parameters or high-bandwidth system support, limiting their application in complex industrial scenarios.

[0005] In addition, in polishing tasks involving flexible materials or fragile components, the demand for contact force control is more stringent. Force overshoot in the contact transient can cause material damage or surface quality degradation, while insufficient contact force can result in suboptimal processing effect. Therefore, how to balance the stability and precision of force control in the contact transient and continuous contact tasks is a core challenge faced by compliant force control technology.

[0006] In recent years, compliant force control algorithms combining optimal control theory and online estimation of environmental parameters have become a research hotspot. For example, environmental stiffness estimation methods based on extended Kalman filter can dynamically adjust control gains during contact to optimize the target interaction force. However, these methods are mostly effective only under static or low-speed contact conditions, and their adaptability to high-dynamic and complex environments still needs to be further improved.

[0007] Based on the above background, the present application proposes a precise compliant force control method for polishing robots, combining adaptive impedance control and optimal control algorithm, aiming to solve the influence of environmental uncertainty on contact force regulation and improve the precision and stability of polishing tasks. SUMMARY

[0008] In order to overcome the above-mentioned defects of the prior art, the present application provides a precise compliant force control method for polishing robots under environmental uncertainty to solve the problems existing in the above background art.

[0009] The present application provides the following technical solutions: a precise compliant force control method for polishing robots under environmental uncertainty, comprising the following steps: Step one, establish an environment-robot dual closed-loop control model according to the set linear environment model and robot dynamics model; Step two, for the unknown part of the environment characteristics in the control model, iteratively solve the optimal control gain in a data-driven manner based on adaptive dynamic programming; Step three, design free space control gain to realize stable relative motion approach to the environment; Step four, design free-contact space control gain smooth transition mechanism.

[0010] Preferably, the step one is specifically: Based on the dynamic interaction relationship between the robot end and the complex environment, an impedance control model considering environmental uncertainty is established; The linear environment model is represented as: ; In the formula, is the environment position, represents the environment velocity, is the environment equilibrium position; the inner loop impedance control defines the damping and stiffness of the environment as and , the contact force between the environment and the end of the robot arm is ; The dynamics model of the robot arm is expressed as: ; wherein, and are the actual end position and the desired end position of the robot arm, represents the actual velocity of the robot, represents the actual acceleration of the robot, is the contact force at the end of the robot arm; represents the stiffness of the robot; represents the mass of the robot; represents the damping of the robot.

[0011] Preferably, the environment-robot arm double closed-loop control model is specifically: A stable environment equilibrium position is defined as , the environment position is equal to the end position of the robot, i.e. , and the force applied on the environment is equal to the contact force applied at the end of the robot arm, i.e. ; wherein, represents the direct contact force between the robot and the environment; A first-order reference model is considered for the instantaneous force response, and is defined as is the first-order reference model reference force: ; wherein, the parameter affects the regulation time of the closed-loop control system, i.e. the bandwidth; is the desired contact force, is the first-order derivative of the first-order reference model reference force, and the reference position from the outer loop control system to the inner loop impedance system is , which is defined as: ; wherein, is the feedback gain of the end velocity of the robot, is the feedback gain of the end position of the robot, is the integral gain of the force tracking error , wherein, , is the feedforward gain of the model reference input force ; represents the integral time.

[0012] Preferably, the environment-robot arm double closed-loop control model is: ; is rewritten as: ; wherein, is the system state, is the system state matrix, is the input matrix, is the system input, is the third derivative of the robot position, is the first derivative of the force tracking error, is the second derivative of the first order reference model reference force.

[0013] Preferably, the adaptive dynamic programming based data-driven iterative solving of the optimal control gain is specifically: The ADP algorithm is used for iterative optimal control gain, and the ADP algorithm based on policy iteration is used for data iterative solving of the optimal control gain; the cost function in optimal control is expressed as: ; wherein, , is the weight of the cost function; represents the integral time; represents a diagonal matrix; The objective function of the optimal control is expressed as: , and is minimum; wherein, represents the optimal control gain of the system.

[0014] Preferably, the specific steps of the ADP algorithm based on policy iteration for data iterative solving of the optimal control gain are: Step a: initialization: given the initial control gain , the initial gain can meet the system stability; Step b: online data collection: let , collect the output state, wherein is a random noise, and calculate the iteration matrix and ; wherein, is related to the system state, the control input, and the control gain, is related to the system state and the weight matrix; Step c: policy evaluation and improvement: the optimal control kernel matrix and the control gain are solved by using the following formula: ; wherein, represents matrix vectorization, represents the next time control gain; Step d: if , then converge, wherein is converged, otherwise let repeat steps c-d until convergence.

[0015] Preferably, the design free-space control gain achieves stable relative motion approach environment, specifically: Based on the impact model of optimization control gain, using the form of relative motion, set the control gain and Zero, feedforward force is set to ,The reference position of the free space is represented as: .

[0016] Preferably, the design free-contact space control gain smooth transition mechanism is specifically: Design control gain smooth switching mechanism, the The first control gain After a time interval Smooth transition from free space control gain To contact space control gain , expressed as: ; Where, Indicates The corresponding filter gain Changes with filter time Value.

[0017] Technical effects and advantages of the present application: The present application is provided with step two, step three and step four, which is conducive to the realization of model-free optimal control based on data iteration by adaptive dynamic programming (ADP), and the impedance effect of force is realized by adopting double closed-loop impedance control, and the control gain can be dynamically adjusted by using contact transient impact modeling, so as to ensure the stability of contact force and processing quality, and the present application is suitable for various complex operation environments, and through the combination of double closed-loop force control framework and optimal control algorithm based on ADP, precise compliant force control of grinding and polishing task can be realized in uncertain environment, the processing precision and stability are effectively improved, and the real-time performance and robustness of the system are also considered. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 The present application is a precise compliant force control method flow chart of grinding and polishing robot in uncertain environment.

[0019] Figure 2 The present application is an intelligent compliant force control method schematic diagram of grinding and polishing mechanical arm and uncertain environment contact process.

[0020] Figure 3The change curve of the nuclear matrix P matrix distance from the theoretically optimal P* matrix when data iteration is performed for the ADP algorithm of the application.

[0021] Figure 4 The change curve of the control gain K matrix distance from the theoretically optimal K* matrix when data iteration is performed for the ADP algorithm of the application.

[0022] Figure 5 The contact force response curve of the force control algorithm of the application in the case of mutation of environmental characteristics.

[0023] Figure 6 The robot position change curve of the force control algorithm of the application in the case of mutation of environmental characteristics. DETAILED DESCRIPTION

[0024] The technical solutions in the application will be described clearly and completely below with reference to the drawings in the application, and additionally, the forms of each structure described in the following embodiments are only examples, and the precision compliant force control method of the polishing robot in an uncertain environment according to the application is not limited to each structure described in the following embodiments. All other embodiments obtained by those skilled in the art without making creative efforts belong to the protection scope of the application.

[0025] As shown in Figures 1-6 , the application provides a precision compliant force control method of a polishing robot in an uncertain environment, which comprises the following steps: Step one, establishing an environment-robot dual closed-loop control model according to a set linear environment model and a robot dynamics model; Step two, based on adaptive dynamic programming, iteratively solving optimal control gain in a data-driven manner for the unknown part of the environmental characteristics in the control model; Step three, designing free space control gain to realize stable relative motion approaching the environment; fully considering the uncertainty of the environmental position; Step four, designing free-contact space control gain smooth transition mechanism; used to ensure that no overshoot force is generated at the moment of contact.

[0026] In this embodiment, it needs to be specifically explained that the step one is specifically: Based on the dynamic interaction relationship between the robot end and the complex environment, an impedance control model considering the uncertainty of the environment is established; The linear environment model is expressed as: ; in the formula, is the environmental position, is the environmental velocity, is the environmental equilibrium position; the inner loop impedance control defines the damping and stiffness of the environment as and , the contact force between the environment and the end of the robot arm is ; The robot arm dynamics model is expressed as: ; wherein, and are the actual end position and the desired end position of the robot arm, represents the actual speed of the robot, represents the actual acceleration of the robot, is the contact force at the end of the robot arm; represents the stiffness of the robot; represents the mass of the robot; represents the damping of the robot.

[0027] In this embodiment, it needs to be specifically pointed out that the environment-robot arm double closed loop control model is specifically: Define a stable environment equilibrium position , the environment position is equal to the end position of the robot, that is, , and the force applied on the environment is equal to the contact force applied on the end of the robot arm, that is, ; wherein, represents the direct contact force between the robot and the environment; Considering the feedforward feedback structure of the outer loop, in order to prevent the expected force from being too large at the moment of input, a first-order reference model is considered for the instantaneous force response, and is defined as the reference force of the first-order reference model: ; wherein, the parameter affects the regulation time of the closed loop control system, that is, the bandwidth; is the expected contact force, is the first-order derivative of the reference force of the first-order reference model, and the reference position from the outer loop control system to the inner loop impedance system is , which is defined as: ; wherein, is the feedback gain of the end speed of the robot, is the feedback gain of the end position of the robot, is the integral gain of the force tracking error , wherein, , is the feedforward gain of the model reference input force ; represents the integral time; The model of the control system of the feedforward-feedback double closed-loop impedance force control, i.e., the environment-robot double closed-loop control model, can be obtained by the above formula as: ; Rewritten as: ; wherein, is a system state, is a system state matrix, is an input matrix, is a system input, is a third derivative of the robot position, is a first derivative of the force tracking error, is a second derivative of the first-order reference model reference force; It is assumed that the robot approaches the environment along the inertial system axis direction, in order to verify the effect of the method, it is assumed that the environment parameters are , , and the dynamic parameters of the robot are , , , the bandwidth of the reference force is .

[0028] In this embodiment, it needs to be specifically pointed out that the optimal control gain is solved by data iteration based on adaptive dynamic programming in a data-driven manner, specifically: The optimal control gain is iterated by collecting input and state data using the ADP algorithm to ensure accurate tracking of the contact force trajectory; The ADP algorithm is used to iterate the optimal control gain, and the cost function in the optimal control is represented as: ; wherein, , is the weight of the cost function; represents the integral time; represents a diagonal matrix; The objective function of the optimal control is represented as: , and is minimum; wherein, represents the optimal control gain of the system; Since the traditional model-based optimal control method cannot be implemented due to environmental uncertainty, the ADP algorithm based on policy iteration is used to iteratively solve the optimal control gain.

[0029] In this embodiment, it needs to be specifically pointed out that the specific steps of using the ADP algorithm based on policy iteration to iteratively solve the optimal control gain are: Step a: initialization: given the initial control gain , initial gain System stability can be satisfied; Step b: online data collection: let , collect output state, where is random noise, and calculate iteration matrix and ; wherein, is related to system state, control input and control gain, is related to system state and weight matrix; Step c: policy evaluation and improvement: solve the optimal control kernel matrix and control gain using the following formula: ; wherein, denotes matrix vectorization, denotes the next time control gain; Step d: if converge, wherein is the convergence threshold, otherwise let , repeat steps c-d until convergence; In this embodiment, the system simulation step is taken as , the weight coefficient , the initial value of the matrix is set to the unit matrix, and since the matrix satisfies Hurwitz, the initial value of the control gain is ; First, according to the known environment model, the theoretical optimal impedance control gain solved based on the model is ; Collect 100 rounds of system data through system input , noise , wherein , the initial state of the system is set to , and the convergence threshold is set to ; through policy evaluation and improvement, the kernel matrix and the control gain converge to the optimal value after only 7 iterations, as shown in Figure 3 , the optimal iteration of the optimal control gain is , which is consistent with the model-based method.

[0030] In this embodiment, it needs to be specifically pointed out that the design of the free space control gain to realize stable relative motion close to the environment is specifically: Optimize the control gain based on the collision impact model to reduce the impact force of the contact transient state and suppress the contact force overshoot; Since the environmental position is unknown, only relative motion form can be adopted, and the contact force is zero in free space motion, which will constantly produce force error, so the control gain is set to zero and The feedforward force is set to to achieve stable feedforward effect, so the reference position of free space is: .

[0031] In the embodiment, it needs to be specifically pointed out that the design of free-contact space control gain smooth transition mechanism is specifically: The control gain smooth switching mechanism is designed to prevent contact force overshoot in the contact moment; In order to ensure smooth transition from free space to contact space, the smoothness of cosine function is used, and the first control gain is changed from free space control gain to contact space control gain : ; wherein, represents the corresponding filtering gain which changes with filtering time ; The effect finally achieved by the ADP-based double closed-loop force control algorithm is shown in Figure 4 and Figure 5 .

[0032] Finally: the above only for the preferred embodiments of the present application, and not for limiting the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application, should be included in the protection scope of the present application.

[0033] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited to this, any skilled person in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.​​​

Claims

1. A method for precise compliance control of a grinding and polishing robot under environmental uncertainty, characterized by: The following steps are involved: Step 1: Establish an environment-manipulator dual closed-loop control model based on the set linear environment model and the manipulator dynamics model; Step 2: For the unknown part of the environmental characteristics in the control model, the optimal control gain is iteratively solved in a data-driven manner based on adaptive dynamic programming; Step 3: Design free space control gain to achieve stable relative motion approach environment; Step 4: Design a smooth transition mechanism for free-contact space control gain.

2. The method for precise compliance control of a polishing robot under environmental uncertainty according to claim 1, characterized in that: The step 1 is specifically as follows: Based on the dynamic interaction between the end of the robotic arm and the complex environment, an impedance control model considering environmental uncertainty is established; The linear environment model is expressed as: Where, is the environmental location, represents the ambient speed, is the equilibrium position of the environment; the inner loop impedance control defines the damping and stiffness of the environment as and , the contact force between the environment and the end of the robot arm is ; The dynamic model of the robotic arm is expressed as: ;in, and are the actual end position and the expected end position of the robot arm, respectively. Indicates the actual speed of the robot. Indicates the actual acceleration of the robot, is the contact force at the end of the robot arm; represents the robot stiffness; Indicates the robot mass; Represents the robot damping.

3. The method for precise compliance control of a polishing robot under environmental uncertainty according to claim 2, characterized in that: The establishment of the environment-manipulator dual closed-loop control model is specifically as follows: Defining a stable environmental equilibrium position , then the environment position is equal to the robot end position, that is , and the force applied to the environment is equal to the contact force applied to the end of the manipulator, that is, ;in, Represents the direct contact force between the robot and the environment; Considering the first-order reference model for the transient force response, define is the first-order reference model reference force: ； Among them, the parameters Affects the adjustment time of the closed-loop control system, that is, the bandwidth; is the expected contact force, is the first-order derivative of the reference force of the first-order reference model, and the reference position input from the outer loop control system to the inner loop impedance system is , defined as: ； in, is the robot terminal speed The feedback gain, The robot end position The feedback gain, Force tracking error The integral gain of , Input forces for model references The feedforward gain of Indicates the integration time.

4. The method for precise compliance control of a polishing robot under environmental uncertainty according to claim 3, characterized in that: The environment-manipulator dual closed-loop control model is: ； Rewritten as: ;in, is the system status, is the system state matrix, is the input matrix, is the system input, is the third-order derivative of the robot position, is the first-order derivative of the force tracking error, is the second-order derivative of the reference force of the first-order reference model.

5. The method for precise compliance control of a polishing robot under environmental uncertainty according to claim 4, characterized in that: The iterative solution of the optimal control gain in a data-driven manner based on adaptive dynamic programming is specifically: The ADP algorithm is used to iterate the optimal control gain, and the ADP algorithm based on policy iteration is used to iterate the data to solve the optimal control gain; the cost function in the optimal control Expressed as: ;in, , is the weight of the cost function; Indicates the integration time; represents a diagonal matrix; The objective function of optimal control is expressed as: ,and minimum; among them, represents the optimal control gain of the system.

6. The method for precise compliance control of a grinding and polishing robot under environmental uncertainty according to claim 5, characterized in that: The specific steps of using the ADP algorithm based on policy iteration to perform data iteration to solve the optimal control gain are: Step a: Initialization: Given the initial control gain , initial gain Can meet system stability; Step b: Online data collection: Let , collect output status, where is random noise, and the iterative matrix is ​​calculated and ;in, Related to the system state, control input and control gain, Related to the system state and weight matrix; Step c: Strategy evaluation and improvement: Use the following formula to solve the optimal control kernel matrix and control gain : ;in, represents matrix column vectorization, Indicates the control gain at the next moment; Step d: If Then it converges, where is the convergence threshold, otherwise , repeat steps cd until convergence.

7. The method for precise compliance control of a polishing robot under environmental uncertainty according to claim 6, characterized in that: The free space control gain is designed to achieve a stable relative motion approach environment as follows: Based on the collision impact model, the control gain is optimized and the relative motion form is used to set the control gain. and is zero, the feedforward force is given by Set to , the reference position in free space is expressed as: 。 8. The method for precise compliance control of a polishing robot under environmental uncertainty according to claim 7, characterized in that: The design of the free-contact space control gain smooth transition mechanism is specifically as follows: Design a control gain smooth switching mechanism, Control gain Elapsed time interval Controlling Gain from Free Space Smooth transition to contact space control gain , which can be expressed as: ;in, express Corresponding filter gain With filtration time The value of the change.

Images